Scientific discovery is bottlenecked by slow manual hypothesis, experiment, and analysis loops

SScientific discovery traditionally follows a slow loop of hypothesis, experimentation, and analysis requiring substantial human effort, even though frontier AI models now reason at PhD level on scientific tasks. The need is for systems that autonomously run complete discovery cycles, proposing candidates, automating lab synthesis and testing, and feeding results back for iterative improvement. This matters for accelerating drug discovery, materials science, and protein engineering.